Antidiabetic prescribing trends and predictors of thiazolidinedione discontinuation following the 2007 rosiglitazone safety alert

This study aimed to examine the change in antidiabetic medication prescribing trends and predictors of thiazolidinedione (TZD) discontinuation six months after the 2007 rosiglitazone safety alert. We performed a retrospective cohort analysis. Patients with two prescriptions for a TZD between 1 January and 21 May 2007, including one covering 21 May 2007, and continuous enrolment during 2006-2007 were identified from the MarketScan database. Multivariate logistic regression analysis was used to compare characteristics between patients who continued and discontinued each TZD. We identified 40,836 and 37,183 individuals with a current prescription for rosiglitazone and pioglitazone, respectively. Significantly more rosiglitazone (53.5%) compared to pioglitazone users (21.4%) discontinued initial therapy six months after the alert (p < 0.001). Approximately 23% of patients who discontinued rosiglitazone were switched to pioglitazone, while <1% was switched from pioglitazone to rosiglitazone. Notably, 19.4% of patients who discontinued rosiglitazone and 36.1% of those who discontinued pioglitazone did not have evidence of any antidiabetic drug at follow-up. There was a significant decrease in metformin and an increase in sitagliptin prescribing in patients who discontinued TZDs. Age, sex, region, cardiovascular comorbidities and physician specialty predicted TZD discontinuation. These findings suggest that FDA advisories may be associated with substantial changes in medication use.